Children with Autism Show No Leftward Asymmetry of Neural Response to Native Vs. Nonnative Speech Distinction

Background: The left hemisphere plays a dominant role in linguistic processing (Zatorre & Gandour, 2008). Leftward lateralization has been found to develop in parallel with children’s advance in linguistic competence (Mills et al., 2004). Similarly, native speakers of a specific language show greater neural sensitivity and efficiency in processing the auditory patterns found in that language (Kuhl, 2010; Zhang et al., 2005). However, such neural specialization appears to be significantly deviated in children with autism. Toddlers with autism failed to show leftward asymmetry when listening to natural speech (Eyler et al., 2012); children with autism show reduced neural sensitivity in discriminating phonological units in the native language but not so much in discriminating non-phonological units and nonspeech contrasts (Huang et al., 2017; Wang et al., 2017; Yu et al., 2015). These findings all point to an altered language-specific neural specialization in children with autism.

Objectives: The current study aimed at examining the neural coding of native vs. non-native phonological processing in children with autism. By comparing native vs. nonnative responses, we expected to investigate the functional specialization in detecting abstract word-level phonological structure of the native language.

Methods: EEG was recorded from 21 school-age (9~13 years) native Chinese-speaking children with autism and 25 age-matched TD controls. The participants were instructed to watch a self-chosen cartoon while ignore any sounds. The native condition consisted of a sequence of Chinese disyllabic nonsense words (e.g., /gu4bo1/), and the nonnative condition included a list of English disyllabic nonsense words (e.g., goober). The stimuli in the two conditions were matched by number of syllables, phonemic units (e.g., CVCV) and key acoustic features (i.e., duration, overall intensity). Mean ERP response amplitude was measured on the left side (F3/C3) and right side (F4/C4) of the scalp, within the time windows of 200-400, 400-600, and 600-800 ms.

Results: Regardless of stimulus and hemisphere, the autism group had smaller response amplitude than the TD group in the two later time windows but not the early window. Both groups showed larger ERP negativity to the native stimuli than to the nonnative stimuli. Specifically, In the 200-400 and 600-800 ms windows, the ERP difference in the TD group was restricted to channels on the left side, whereas the difference in the autism group was bilaterally distributed. In the 400-600 ms window, the ERP difference was mainly driven by the autism group with the effect only on the right side.

Conclusions: Both children with autism and TD children differentiate the abstract phonological structures of native vs. nonnative speech at the neural level. The ERP differences in the TD children show a clear leftward distribution, which coincides with existing adult data that linguistic information in native speech can engage specialized neural network for language (Perrachione et al., 2009). In comparison, the lack of leftward asymmetry in autism might represent a neural signature of impaired or delayed neural specialization for native language processing. Data from nonspeech stimuli are needed to examine the role of prosody in such language-specific neural specialization.